University of Minnesota
Dispersed Data-driven Computing

Instructional Cloud for Class Projects

An instruction cloud (running VMs) has been set up for use in course projects as needed, and will be shared by the project teams that decide to use it. This cloud will have remote access from CS and CSE machines (or from outside using CSE VPN).

The instructional cloud consists of multiple VMWare-based virtual machines running on each physical host. Each group that wants to use this instructional cloud will be assigned a set of VMs (a "private virtual cluster") with full root access, so they can deploy any software they need. Each virtual cluster will be firewalled and will have Hadoop/HDFS set up on it as a start.

Since the class has no TA, there is limited systems support available, so that the project groups using the cloud will be responsible for installing, configuring, and maintaining any software and tools needed for their purposes. You can choose to use it based on your project needs and appropriateness to your project goals. If you plan to use the cloud, but need additional VMs, you will need to talk to the instructor about your needs, so that your private virtual cluster could be expanded or the cloud access could be suitably multiplexed between multiple groups. In addition, even though each group will be given a private virtual cluster, during the course of the semester, you may have to coordinate your cluster usage with other groups to avoid interfering with each other (e.g., if your experiments are sensitive to other background load or traffic on the cloud or network). Finally, the cloud is not backed up, so make sure to transfer any code/data you need to save to the students' home directories via scp.